Research And Implementation Of Face Recognition System For Mobile Phone Unlocking Based On Android Platform

Recently, with the rapid development of Internet technology and mobile intelligent terminal, more and more applications appear on the mobile phone that involving personal privacy, so that the information security issues have increased. Face recognition is a kind of biological recognition methods with the advantages of natural, convenience and non-contact. The technology has become a hot research topic in the field of personal authentication technologies. The user's face image captured by the face recognition system is often affected by environmental conditions, such as the varying illumination, complex background and so on. Besides that, the system platform of the mobile phone also has restrictions of the storage capacity, speed of operation and other aspects.(1) Research of face detection algorithm. This paper studies the combination algorithm based on skin color detection and the Adaboost face detection. This algorithm first establishes the skin color filter to reduce the effects of illumination.Then the Adaboost algorithm is used to train and select the weak classifier and the strong classifier. Finally, this method can detect faces in the environment including varying illumination and complex background with multiple Adaboost classifier cascade. Experiments show that this method has a better effect.(2) Research of facial feature extraction algorithm. This paper studies a method which combines the PCA and LDA and it obtains a better output. But this algorithm can't solve the nondeterminacy of the membership for each class caused by the illumination or expression. So the method is studied which introduces the fuzzy technology. This method extracts initial feature by using PCA and then computes the value of the membership for each class, finally LDA is employed for further extraction. Experiments show that this method has a better extraction effect.(3) Research of face recognition algorithm. A method called NN-SVM is studied which combines nearest neighbor(NN) and support vector machine(SVM). This algorithm uses NN to classify for the first time and then uses SVM to do a more precise classification. Experiments show that the algorithm can improve the accuracy of classification better compared to sole NN or sole SVM.(4) Design and implementation of face recognition system for mobile phone unlocking. The system requirement analysis is performed. On this basis, the system is created according software engineering after learning the architecture of the Android system. The system frame is built on Android platform, and Java is used to write the core modules which include face detection, feature extraction, face recognition and mobile phone unlocking for the implementation of face recognition system for mobile phone unlocking.